why info of h264 svc temporal layer in webrtc doesn't match description in RFC6190?

[13000...@qq.com]

RFC6190 makes a specification on Scalable Video Coding,  in section 1.1.3 of RFC6190,  it mentions that using a NAL unit header extension to support SVC support(format shown as below)

```

            +---------------+---------------+---------------+
            |0|1|2|3|4|5|6|7|0|1|2|3|4|5|6|7|0|1|2|3|4|5|6|7|
            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
            |R|I|   PRID    |N| DID |  QID  | TID |U|D|O| RR|
            +---------------+---------------+---------------+

```

it seems that 'tid' represent the temporal layer of current NAL,  to verify this,  I use windows to compile and run webrtc; In H264EncoderImpl::InitEncode( ),  I chagne the code for support enable svc temporal layer

```

//h264_encoder_impl.cc

int32_t H264EncoderImpl::InitEncode(const VideoCodec* inst,

                                    const VideoEncoder::Settings& settings) {

.....

    //configurations_[i].num_temporal_layers = codec_.simulcastStream[idx].numberOfTemporalLayers;

    configurations_[i].num_temporal_layers = 3;  //enable svc temporal layer to 3

.....

}

```

Since there a object named 'info'  which includes the infomation of temporal layer after frame encode, so I can compare the value of info.temporal_layer with value of 'tid' to validate；so I add a judgement and breakpoint as below (code with highlight)

```

//h264_encoder_impl.cc

static void RtpFragmentize(EncodedImage* encoded_image,

                           const VideoFrameBuffer& frame_buffer,

                           SFrameBSInfo* info,

                           RTPFragmentationHeader* frag_header) {

         .....

          for (int layer = 0; layer < info->iLayerNum; ++layer) {

          const SLayerBSInfo& layerInfo = info->sLayerInfo[layer];

          // Iterate NAL units making up this layer, noting fragments.

          size_t layer_len = 0;

          for (int nal = 0; nal < layerInfo.iNalCount; ++nal, ++frag) {

                  // Because the sum of all layer lengths, |required_capacity|, fits in a

                  // |size_t|, we know that any indices in-between will not overflow.

                  RTC_DCHECK_GE(layerInfo.pNalLengthInByte[nal], 4);

                  RTC_DCHECK_EQ(layerInfo.pBsBuf[layer_len + 0], start_code[0]);

                  RTC_DCHECK_EQ(layerInfo.pBsBuf[layer_len + 1], start_code[1]);

                  RTC_DCHECK_EQ(layerInfo.pBsBuf[layer_len + 2], start_code[2]);

                  RTC_DCHECK_EQ(layerInfo.pBsBuf[layer_len + 3], start_code[3]);

                  frag_header->fragmentationOffset[frag] =

                        encoded_image->size() + layer_len + sizeof(start_code);

                  frag_header->fragmentationLength[frag] =

                layerInfo.pNalLengthInByte[nal] - sizeof(start_code);

                  layer_len += layerInfo.pNalLengthInByte[nal];

          }

          // Copy the entire layer's data (including start codes).

          memcpy(encoded_image->data() + encoded_image->size(), layerInfo.pBsBuf,

                 layer_len);

          encoded_image->set_size(encoded_image->size() + layer_len);

          // add a judgement and breakpoint

          // after frmae encode, try to compare info.temporal_layer with tid

          RTC_LOG(INFO) << "xxxxxxxxxxxxxxxxxxxxxxxxxxxx temporal layer is " << layerInfo.uiTemporalId;

          if (layerInfo.uiTemporalId != 0) {

            int a;

=>        a = 1;  // add breakpoint here

          }

          .....

}

.....

}

```

then brakpoint was triggered,  log says temporal layer is 2 :

```

(h264_encoder_impl.cc:150): xxxxxxxxxxxxxxxxxxxxxxxxxxxx temporal layer is 2

```

but value 'tid' in nal is 0, anything wrong?

[Alexandre GOUAILLARD]

I don't think libwebrtc supports H264 SVC.

At best, it would support what google calls k-SVC, which is a mix of simulcast (for spatial resolutions) and SVC (only for temporal resolutions).

--

---
You received this message because you are subscribed to the Google Groups "discuss-webrtc" group.
To unsubscribe from this group and stop receiving emails from it, send an email to discuss-webrt...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/discuss-webrtc/98c01e52-681c-4aef-90e9-c217b13b4a7f%40googlegroups.com.

--

Alex. Gouaillard, PhD, PhD, MBA

------------------------------------------------------------------------------------

President - CoSMo Software Consulting, Singapore

------------------------------------------------------------------------------------

sg.linkedin.com/agouaillard

• 
  

[Sergey Silkin]

WebRTC uses frame marking RTP extension [1] for signaling temporal layer [2].

[1] https://tools.ietf.org/id/draft-ietf-avtext-framemarking-05.html

[2] https://cs.chromium.org/chromium/src/third_party/webrtc/api/video/video_frame_marking.h?rcl=f2dc05978f1dda877d811b81d1ee9cedb6d38be1&l=22

To unsubscribe from this group and stop receiving emails from it, send an email to discuss-webrtc+unsubscribe@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/discuss-webrtc/98c01e52-681c-4aef-90e9-c217b13b4a7f%40googlegroups.com.